Multi-Source Performance Appraisal: An Empirical and Methodological Note

Performance appraisal is an important issue in the area of Human Resource Management (HRM). There is considerable agreement as to the centrality and importance of the performance appraisal process among scholars concerning both public (Heneman et al., 1989; French, 1987) as well as private organizations (Foltz et al., 1983; Lacho et al., 1991). We may learn of the importance attached to the performance appraisal system through the large number of organizations that use various kinds of formal performance appraisal procedures--about 90% of all organizations (Lazer and Wikstrom, 1977; Varney, 1972). There is a consensus that performance appraisal is a key process which is vital to the existence and development of the organization (Latham and Wexley, 1981; Fombrun and Laud, 1983). The term assessment is associated with a variety of concepts, programs, and procedures. It includes measurement, evaluation, correlation, and prediction with one or more variables and criteria, as well as standard of performance (Herrold, 1986). Illustratively, Milkovich and Boudreau (1988) defined performance appraisal as the process that measures employee performance, which involves four decisions: (1) what to assess; (2) who should make the assessment; (3) which assessment procedures to use; and (4) how to communicate assessment results. This article focuses on the third decision and presents an innovative approach that combines information from several sources in order to reach more valid and reliable performance appraisal (Latham and Wexley, 1981; Schneier, 1977; Tsui, 1984; Tsui and Ohlott, 1988). This approach may contribute to a reduction in performance appraisal problems which are usually due to poor information from one or more of the performance appraisal sources. The authors will not consider other appraisal problems such as inadequate descriptions of criteria or difficulties in methods. There are several objectives in employing a performance appraisal system. The main one is to satisfy the need of the organization for information about its employees' abilities, qualifications, training needs, and potentials. Another objective is to provide feedback to the workers. In order to serve all of these needs, the performance appraisal system must be valid and reliable. There are several ways to improve the performance appraisal system. The common recommendation for improvement of performance appraisal systems is to base them on behavioral scales instead on traits evaluation (Murphy and Cleveland, 1991; Heneman et al., 1984). Another recommended method is to base the appraisals on more than one source of information. There are, however, difficult problems in developing such a composite model and most organizations use only one source of information--the direct supervisor, as reported in Field and Holley (1977) and in Lazar and Wikstrom (1977). This problem limits the feasibility of implementing and evaluating the effectiveness of a composite criteria model. Literature review uncovered several sources of information available for appraising pubic sector employees, with the most common ones presented in Figure 1 (Ammons, 1987). (Figure 1 omitted) The main one is the direct manager's appraisal of his or her subordinates. Other possible sources are subordinate's appraisal or self-appraisal (Steele, 1985), objective information, and sources outside the organization (such as independent advisors or customers, suppliers, etc.). All relevant sources have shortcomings affecting the validity and reliability (see Bernandin, 1986; Ashford, 1989). In addition, it is difficult to ascertain the relative validity and reliability of the sources. Some researchers have recommended that the quality of performance appraisal could be improved by combining sources but none suggest a method for accomplishing this task (Schneider and Carrol, 1977; Miner and Miner, 1985; Kavanagh, 1982). Borman (1980), as cited in Kavanagh (1982) argued that, although it is possible to gain more information, accuracy may not increase when multiple sources are used. …